Experimental Determination of Hydrogen-Air Laminar Burning Velocities, and the Effect of Water Mist

(1)

Experimental Determination of Hydrogen-Air Laminar Burning

Velocities, and the Effect of Water Mist

Wulme Puoru Dery

A thesis in partial fulfilment of the requirements for the degree of Master of Science in the subject of Physics Process Safety

Technology

University of Bergen, Department of Physics and Technology

(2)

Abstract

This research project characterises an experimental set-up with the aim of finding laminar burning velocities of premixed hydrogen-air (dry) mixtures, at initial pressure and pressure of 294 K and 1 bar, using a 20 litre cubic explosion chamber.

Two different methods were employed; a constant-pressure method (CPM) applied with a schlieren imaging technique, and a constant-volume method (CVM) adopting the pressure-time history. Regarding the former method, linear and non-linear relation between the propagation of flame and stretch rate was employed to obtain the unstretched flame and burning velocity.

Using the pressure measurements during the transient dispersion process, CPM using a pressure transducer was employed to calculate the burning velocity, relating the pressure rise to the radius of the flame.

Both measuring techniques gave laminar burning velocities in agreement, with a maximum relative difference of 10 %. Observed wrinkling in the flame propagation due to hydrodynamic instabilities, in addition to limitations associated with experiments performed in a cubical vessel, made it difficult to produce quantitative results unison with those found in other literature.

Using a nozzle, hydrogen-air mixtures were introduced to water mist with concentrations between 0.08-0.3 m³/kg. This was done to investigate its influence on the laminar burning velocity. The inclusion of water mist gave burning velocities increased with as high as 100 %, compared to the dry results. This was mainly due to the generation of turbulence during the injection of water mist.

(3)

Preface

This is a Master’s thesis submitted as part of a Master of Science degree in the subject of Process Safety Technology at the Department of Physics and Technology (DPT) at the University of Bergen. The experimental work was carried out in the Bjørn Trumpy House belonging to DPT, and in cooperation with the company Gexcon AS. The work started in the fall of 2016 and was completed in June 2017.

(4)

Acknowledgements

First of all, I would like to thank the University of Bergen for enabling me the opportunity to conduct an experimental research thesis, and also to Gexcon AS for the cooperation in implementing it. As with most research projects, obstacles along the way is presumably inevitable. Fortunately, I have had many able minds to turn to along the way:

My supervisors Bjørn Arntzen and Morten Holme for excellent guidance, and enduring feedback during the entire process, Kari Halland for initially showing me around the lab, Werner Olsen for assistance in making and understanding the electronic equipment, Roald Langøen and Charles Sebastiampillai for exceptional work done in the mechanical workshop constructing and modifying apparatus used, Rachid Maad for helping install the LabVIEW-software needed, Yi-Chun Chen for help understanding the optical system and lending me lenses, Laurence Benhard for assisting in fine-tuning the optical system and Andre Vagner Gaathaug for welcoming me to observe their schlieren set-up at the University College in Telemark.

My fellow classmates with whom I shared an office with, and otherwise en- countered regularly during this period has been immensely cherished. The comradely atmosphere, may it be during lunch (s/o Tran Hieu Thi) or playing fooseball will not be forgotten, and friendships made assuredly preserved. I’d also like to thank those along the way who provided dank memes (Philippe, Andre, Sindre), supplied with good music (Ebro), and proofread (Jon, Tone, Marcus). Last but not least, I’d like to thank my parents for endless love, support and encouragement, and my brother for just being there.

“I think it’s possible to learn. The problem is that we learn so damned slowly, so that by the time you’ve realized something, it’s too late.” - Harry Hole

(5)

List of Figures

2.1 One-dimensional combustion model with premixed fuel gas-air burning at constant pressure with a plane, laminar flame in a tube. . . 10 2.2 Ideal laminar spherical burning and expansion of a premixed

gas-air mixture, after ignition at midpoint. . . 11 2.3 Diagram of light rays entering a new material and being re-

fracted in proportion to a refractive-index gradient. . . 17 2.4 Z-type schlieren arrangement. . . 19 2.5 Schlieren image showing density gradients in the test area after

a hydrogen-air explosion. . . 20 3.1 The comparison of laminar flame velocities of hydrogen-air

mixtures measured by various techniques. . . 23 3.2 Ignition and propagation of hydrogen-air exposed to water

mist flame fed through a nozzle initially at 305 K and 1 bar without(a) and with(b) water mist. . . 24 3.3 Evaporation time as a function of droplet diameter for various

relative hot gas flow speeds around the droplet. . . 25 4.1 20-litre cubical explosion chamber. . . 27 4.2 From left to right - Piezoelectric pressure sensor (Kistler

7031); Mounting adapter (Kistler 7501); BNC-mounting nipple (Kistler 7401). . . 28 4.3 Light, lens and slit aligned accordingly. . . 31 4.4 Illustration of the effect of the knife-edge movement on the

schlieren image for a vertical knife edge cutting parts of the light beam. . . 32 4.5 Data acquisition module, NI USB-6259, used for equipment-

triggering and acquisition of amplified pressure signals in pressure measurements. . . 33

(9)

4.6 The mounted water cylinder and manometer connected to the explosion chamber. . . 35 4.7 Circuit diagram and photograph of the pulse amplifier. . . 36 4.8 Sketch of the experimental equipment and set-up used, exclud-

ing the optical system. . . 37 5.1 Schlieren and corresponding binary image. . . 41 5.2 Cropped binary image just under the electrodes to extract the

diameter measured in pixels. . . 41 5.3 Calculating the pixel-to-centimetre ratio, using a binary image

of a calliper with a known distance between the pivoted legs of 1.150 cm. . . 42 5.4 Flame radius with time after ignition within the valid range of

extrapolation, showing a third-order polynomial equation. . . 44 5.5 Propagating flame velocities plotted over the stretch rate curve

and a fitted linear extrapolation. . . 45 5.6 Propagating flame velocities plotted over the stretch rate curve

and a fitted non-linear extrapolation. . . 46 5.7 Various initial and equilibrium state calculated results by CEC. 47 5.8 Internal chamber pressure-time history data for a hydrogen-air

mixture, showing both raw and filtered pressure data. . . 48 5.9 The laminar burning velocity obtained as a function of pressure,

and the extrapolation back to initial pressure, Su0. . . 50 6.1 An expanding flame image taken, using a plano-cylindrical

lens between the light source and the first mirror to reduce astigmatism. A fuzzy flame edge is exposed. . . 53 6.2 Captured sequence of schlieren images showing expanding flame

propagation for equivalence ratios φ = 0.7, 1.3, 1.9 from the first row downwards respectively. . . 55 6.3 Extracted flame front radius from schlieren optical images. . . 56 6.4 Calculated stretch rates as a function of flame speed, with

linear extrapolation for extracting the unstretched flame speed and Markstein number. . . 57 6.5 Non-linear extrapolation for extracting the unstretched flame

speed and Markstein number. Extrapolation is shown as the dashed line. . . 57 6.6 Expansion ratio for various hydrogen-air mixtures obtained

using CEC. . . 58 6.7 Laminar burning velocities for hydrogen-air mixtures, measured

at initial temperatures of 294 K and atmospheric pressure. . . 59

(10)

6.8 Pressure-time history for all seven hydrogen-air mixtures during the CVM combustion event, showing the raw and corresponding filtered data. . . 60 6.9 Measured maximum explosion pressure for a hydrogen-air mix-

tures using a 20 and 120 litre vessel, used in the work in this thesis and in the work of Cashdollar et at. respectively, compared to the calculated equilibrium pressure. . . 61 6.10 Calculated flame radius rate as a function of time, obtained

from the pressure-time history. . . 62 6.11 Calculated and extrapolated laminar burning velocities as a

function of the dynamic pressure inside the vessel for various mixtures. . . 62 6.12 Laminar burning velocity for hydrogen-air mixtures measured

at initial temperature (294 K) and atmospheric pressure (1 bar) for each of the methodologies applied. . . 63 6.13 Comparison of laminar burning velocities measured in this work,

and by other authors using various techniques in hydrogen-air mixtures at atmospheric pressure and room temperature. . . . 64 6.14 A schlieren image taken 0.5 seconds after water mist has been

injected into the chamber, with the visibility already severely compromised. . . 65 6.15 Using CEC, the calculated maximum pressure for water mist

present in the initial hydrogen-air mixture are obtained with varying water mist concentrations. . . 66 6.16 The experimentally found maximum pressures for water mist

present in the initial hydrogen-air mixture, with varying water mist concentrations. . . 67 6.17 Comparison of estimated burning velocity measured with the

introduction of water mist with various concentrations. . . 68 6.18 Tabulated SMD distributions generated through pressure at-

omization, provided by the manufacturer Hansa Engineering AS. . . 69 6.19 Flow rate injected into the chamber, found from the pressure

(p_nozzle) duration needed for the highest concentration of water mist (0.3 kg/m³). . . 70 6.20 Pie charts displaying the injected water droplet SMD distribu-

tion, based on the dynamic pressure injected, p_nozzle. . . 71

(11)

6.21 Burning velocities of hydrogen-air mixture (φ = 1.1), showing the effect of varying the time between water mist injection (0.30 kg/m³ H2O) and moment of ignition. The dashed line represents the burning velocity obtained without water mist present. . . 73 6.22 Comparison of estimated burning velocities of various amount

of water mist, and ignition with and without 40 s latency. . . . 74 6.23 The flame and pressure distribution, non-uniform in a cubic

enclosure and uniform in a spherical enclosure. . . 75 6.24 Varying the lower pressure range of measured pressure used to

calculate and extrapolate the laminar burning velocity, at φ = 1.5. . . 77 6.25 Different extrapolation ranges for the measured pressure (p₀

= 1.5, 2, 2.5) used in calculating S_u₀ compared to laminar burning velocities measured in this work and in other literature. 78 6.26 Air pressure (hPa) measured at Florida, Bergen between 01.03.2017

- 07.03.2017. . . 79 6.27 Variation in the expansion ratio and burning velocity with

initial temperature. . . 80 6.28 Water mist trajectory from the water nozzle at the outset. . . 81 B.1 Onset of water mist behaviour, 10 ms after injection with a

flow rate of ≈0.031 l/min. . . B.2 B.2 A series of photograph showing the different latency effects of

water mist behaviour in a cubical vessel after injection. . . B.3

(12)

List of Tables

2.1 Combustion system ordered with respect to premixedness and

flow type. . . 6

2.2 Mixtures described by equivalence ratio. . . 7

2.3 Flammability limits for various gasses. . . 8

2.4 Specific heat capacity coefficients for various gases. . . 9

3.1 Data by various authors showing the hydrogen-air premixed laminar burning velocity measured at stoichiometry, S_u^st, and maximum, S_u^max, using different methods. . . 22

6.1 A quality control of the pressure transducer using a manometer as a secondary pressure source. . . 54

6.2 Assummed and calculated water mist concentrations based on the duration of p_noxxle, injecting water mist into the explosion chamber. . . 70 A.1 Initial dry tests. . . A.1 A.2 Tests with no latency between water injection and ignition,

with 0.30 kg/m³ water mist concentration. . . A.2 A.3 Tests with no latency between water injection and ignition,

with 0.18 kg/m³ water mist concentration. . . A.2 A.4 Tests with no latency between water injection and ignition,

with 0.08 kg/m³ water mist concentration. . . A.2 A.5 Tests with various latencies between water injection and igni-

tion, with 0.30 kg/m³ water mist concentration. . . A.3 A.6 Tests with 40 s latency between water injection and ignition,

with 0.08 kg/m³ water mist concentration. . . A.3 A.7 Tests with 40 s latency between water injection and ignition,

with 0.18 kg/m³ water mist concentration. . . A.3 A.8 Tests with 40 s latency between water injection and ignition,

with 0.30 kg/m³ water mist concentration. . . A.4

(13)

Nomenclature

Abbreviations

BNC Bayonet Neill-Concelman

CEC Chemical Equilibrium Calculator CPM Constant-Pressure Method

CVM Constant-Volume Method

DPT Department of Physics and Technology FLACS FLame ACceleration Simulator

FPS Frames Per Second

LFL Lower Flammability Limit

PFV Photron Fastcam Viewer

SMD Sauter Mean Diameter

UFL Upper Flammability Limit

UoB University of Bergen

Other Symbols

φ Equivalence ratio

f /number Focal ratio

n_air Mole fraction of air in mixture n_{f uel} Mole fraction of fuel in mixture

vol% Volume percent

vol_{f uel} Volume fraction of fuel in mixture vol_total Total volume of gas mixture Latin and Greek Symbols

α Stretch rate 1/s

γ Specific heat ratio

ρ_b Density of burned gas kg/m³

ρ_u Density of unburned gas kg/m³

ξ Light refraction angle

A Flame surface area m²

C Capacitance F

c Speed of light in vacuum m/s

(14)

C_p Specific heat capacity at constant pressure J/(kg K) C_v Specific heat capacity at constant volume J/(kg K)

D Diameter mm

E Energy stored in a capacitive spark J

E Expansion ratio

f Focal length mm

K K-factor l/(min bar)

k Gladstone-Dale coefficient cm³/g

L Markstein length mm

n Number of moles

n Refractive index

p Pressure bar

p_i Initial pressure bar

p_max,ad Adiabatic isochoric equilibrium combustion pressure bar

p_max Maximum explosion pressure bar

p_nozzle Water pressure on the nozzle bar

R Gas constant J/(K mol)

r_f Flame radius m

R_s Specific gas constant J/(K kg)

R_v Radius of the vessel m

S_f Flame speed m/s

S_g Unburned gas flow velocity m/s

S_u Laminar burning velocity m/s

S_u₀ Burning velocity at initial reference conditions m/s

T Temperature K

t Time s

U Voltage V

V Volume m³

v Light speed in a medium m/s

(15)

Chapter 1 Introduction

1.1 Motivation

The leading objective of this thesis is to experimentally obtain laminar burning velocities, a fundamental input parameter necessary to simulate dispersion and/or gas explosions, with particular spotlight on scenarios whereby water mist is present.

The aim, together with the Safety Technology subgroup within the Process Technology group at the Department of Physics and Technology (DPT), is to focus on safety technology and combustion phenomena. To acquire research-based knowledge about various parameters related to combustion and explosion hazards especially on gas and dust explosion [1]. Both experimentally and numerically, research tasks are often collaborated with external agencies, such as Gexcon, one of the leading research environments in the world in the field of gas explosions. To investigate parameters associated with explosions, Gexcon carries out safety assessments, research projects and physical testing of gases and dusts including part taken in investigations of large accidents like Piper Alpha in 1988 and the Texas City refinery in 2005 [2].

Gexcon develops, maintains and uses the industry standard software for modelling gas explosions; FLame ACceleration Simulator (FLACS). As a computational fluid dynamics tool, the software can be used in full 3D for all typical flammable and toxic release scenarios predicting consequences much more accurately and including effects of all contributing and mitigating effects such as confinement and congestion due to existent geometry, ventilation

(16)

and deluge [2]. The software became commercially available in the 1990s, and to meet the requests from the industry, continuous development and maintenance is necessary. In cases whereby little to no accredited literature data is available for input parameters for various gases in the FLACS software relating to combustion, an effort to locally obtain these values is applied. The laminar burning velocity in particular, is an important mixture property and is extensively used in extracting information about diffusivity and reactivity for any given fuel-air mixture [3].

With regards to this, in recent years the term "hydrogen economy" has experienced a spike in attention due to its potential as an energy carrier, transitioning the use of fossil-fuel-based energy to a more sustainable one [4].

Hydrogen, despite being one of the most common chemical element, does not exist in nature in its pure form. It has to be separated from hydrocarbons or other hydrogen carriers, either by electrolysis from water or by chemical processes. Both these instances of hydrogen formation may occur in a nuclear power plant, either as part of normal operations, or following a severe accident [5]. Hydrogen has a wide flammable range when it is mixed with air (4 - 74 vol%) and once uncontrolled dispersion has occurred, the likelihood of local hydrogen concentration materializing in various zones of a containment building is high. The eventual outcome once an ignition source is present is deflagration or even more dangerously - an explosion. Especially in plant situations in the nuclear industry, an overpressure following an explosion could threaten the integrity of the containment and impose radioactive pollution.

In relation to nuclear safety and a hypothetical severe accident, much work has been, and is still conducted to investigate possible mitigations, with ventilation and inerting techniques being the two main paths usually taken [6,7]. Ventilation involves providing frail areas to the confinement, preventing the pressure rise during an explosion to be confined. The hot expanding gas products and unburned fuel may escape through the weakened areas, which are designed to open once a pre-specified pressure level is reached. In nuclear plants however, requirements such as the vent size and corresponding location might be too impractical to implement for it to successfully mitigate an explosion. Inerting on the other hand works by adding a non-reactive agent such as nitrogen, carbon-dioxide or argon to diminish the flammable mixture to an adequately low level, preventing flame propagation and a possible explosion.

It is however imperative that the inert gas is able to infiltrate the whole region in the enclosure to effectively suppress the flammable concentration, which may be a challenge in a building with complex geometry.

Even though it might not be capable of rendering a flammable mixture full

(17)

inert practically, steam and water mist added to a hydrogen-air mixture are two other mitigation techniques explored [8,9]. Being ubiquitous, non-toxic, environmentally favourable and with a high heat capacity per unit mass, water has always been regarded as an ideal fire suppression agent [10]. Water mist (99% of the droplets are less than a millimetre in diameter) in particular can have a significant mitigating effect upon the burning velocity and ignition behaviour of any type of flame. With an exposed surface area depending on the droplet diameter, it reduces the rate of reaction by extracting heat from the flame, both as a liquid component and after vaporisation as a diluent [11].

Experimental studies on the interaction between hydrogen-air laminar burning velocities and water mist is currently still limited. The presence of dense water mist makes obtaining the laminar burning velocity a bit more intricate.

Customarily, the burning velocity is obtained directly from the examination of flame; either by curved flames in stagnation flow, propagating spherical flames in combustion vessel, flat flames stabilized on burner or conical flames stabilized on a bunsen burner [12]. However, these methods require a clear and translucent field of vision, something a dense water mist does not apply to. It may also generate strong flame instabilities, making it harder to obtain a well-defined flame front for measure purposes.

Other methods can be employed in obtaining the laminar burning velocity that is considered less convoluted without direct observation of the flame, when more advanced means are unattainable. In a confined explosion vessel, an indirect method of obtaining a value of the flame propagation through a pressure-time history can determine the corresponding burning rate. In a constant volume vessel and assuming that the fractional pressure rise is proportional to the fractional mass burned, a relationship between the measured rate of pressure and the burning velocity can be found [13,14].

(18)

1.2 Objectives

The objective was to develop and improve on an experimental rig whereby premixed hydrogen-air mixture was ignited. The spherically expanding flames generated were to be extracted using two methods; by a z-type schlieren imaging system and by a pressure-time history method with the latter method considered less inferior in accuracy. Both methods used in obtaining the laminar burning velocity, against each other and literature data was the ultimate deciding factor determining their validity. Finally, an introduction of water mist into hydrogen-air mixture was evaluated to see how it influenced the laminar burning velocity.

Prior to the work in this thesis, all major apparatuses needed to conduct this study was procured though the work of Halland [15] in 2015, conducting her thesis project at UoB. This included an explosion chamber, a high-speed camera, parabolic mirrors and pressure transducers.

(19)

Chapter 2 Theory

2.1 Basic concepts

This chapter describes the basic theory, with the intention of highlighting important concepts essential to the work in this thesis. Basic in the sense that the reader, already familiar with the contents or not, may reach a common fundamental perspective accordingly.

2.1.1 Combustion

In combustion processes, fuel and oxidizer (typically oxygen in air), are mixed and burned. As a phenomena, combustion is found difficult to define precisely and there exists many attempts of a formal definition. A general definition found in the literature is given below iterated by Williams [16]:

"Combustion may be considered to be the science of exothermic chemical reactions in flows with heat and mass transfer."

Combustion is usually accompanied by the generation of heat and emission of light in the form of a flame, a self-sustaining propagation of a localized combustion zone at subsonic velocities [17]. Flame inhibits several categories based upon whether the fuel and oxidizer is mixed first and burned later, or whether combustion occurs simultaneously with the mixing, premixed and non-premixed respectively.

Tab. 2.1 illustrates examples of combustion systems that belong to each of these categories including the flow conditions. Fluid motion is conditioned

(20)

as either laminar or turbulent combustion with laminar fluid moving with low velocities lacking cross-currents or eddies [17]. Meanwhile turbulent combustion is characterized by higher velocities with the fluid moving more erratically and lateral mixing forms taking place.

Table 2.1: Combustion system ordered with respect to premixedness and flow type.

Fuel/Oxidizer Fluid Motion Examples

Premixed Turbulent Explosion

Laminar Bunsen flame

Non-premixed Turbulent Pulverized coal combustion

Laminar Candle

For the work of this thesis a premixed laminar flame is inhibited, as the fuel and oxidizer are mixed and ignited.

2.1.2 Explosions

Bang! The noise whereby a rapid increase in pressure has occurred is usually associated with an explosion. By a definition suggested by Eckhoff [18], an explosion is a exothermic process that, when occurring at constant volume, gives rise to a sudden and significant pressure rise. The energy release by the exothermic process can either be physical (e.g. bursting of a pressurized vessel), chemical (e.g. rapid combustion) or nuclear (fusion or fission) [17].

Disregarding explosives and chemically unstable substances, there are five requirements necessary for a gas explosions to occur:

1. Fuel: Combustible gas, vapour or dust

2. Oxidizer: Usually oxygen in air, but not limited to (as in the case of explosives)

3. Combustible mixture: Proper dispersion and concentrations for combustion

4. Confinement: Not a necessity for an explosion, but its impact on the pressure built up is extensive

5. Ignition source: Any heat source capable or initiating an exothermic chain reaction

(21)

2.1.3 Equivalence ratio

The equivalence ratio describes the concentration of fuel in the oxidizer and controls the combustion process as in a complete combustion there is sufficient oxidizer for all the fuel to react [17]. Defined as the the ratio of the fuel-to-oxidizer ratio to the stoichiometric fuel-to-oxidizer ratio:

φ= _n

f uel

nair

mixture

_n

f uel

nair

stoichiometric

. (2.1)

If there is an excess of fuel, the system is fuel-rich, and if there is an excess of oxygen it is referred to as fuel-lean. Accordingly, as shown in Tab. 2.2, premixed combustion processes can be divided into three groups:

Table 2.2: Mixtures described by equivalence ratio.

Mixture φ

Lean < 1

Stoichiometric 1

Rich > 1

The balanced chemical equation for a hydrogen-air mixture and a hydrogen- air-water mixture at equilibrium is given below as Eq. (2.2) and Eq. (2.3).

It has been taken to account that dry air contains only about 21% oxygen and 79% nitrogen. Thus for each oxygen molecule, there are 3.762 nitrogen molecules.

H₂+ 0.5 (O₂+ 3.762) N₂ −−→H₂O + 0.5·3.762 N₂. (2.2) H₂+ 0.5 (O₂ + 3.762) N₂+ H₂O−−→2 H₂O + 0.5·3.762 N₂. (2.3) In terms of volume percentage, the amount of volume fuel constituted as part of the total volume, can be used to describe the gas concentration:

vol% = vol_{f uel}

vol_total ·100, (2.4)

where vol_{f uel} represents the fuel-volume and vol_total the total volume of the

(22)

The reactants and the products of the combustion can be considered asideal gases, and the equation of state is applicable as follows:

pV =nRT, (2.5)

wherep is pressure, V is volume, n is number of moles, R is the universal gas constant and T is temperature [17].

Rearranging Eq. (2.5), the pressure can be expressed asp=n^RT_V . At constant volume and temperature it can be stated that the total pressure of a gas mixture is determined by the total number of moles of gas present:

vol_{f uel} voltotal

= p_{f uel} ptotal

= n_{f uel} ntotal

. (2.6)

2.1.4 Flammability limits

As the fuel to air ratio is either increased or decreased, there are limits to which the mixture no longer is able to propagate a flame. These two finite upper and lower flammability limits, UFL and LFL respectively, are defined by experimental determination and differ for various gasses as shown in Tab.

2.3 [19].

Table 2.3: Flammability limits for various gasses.

Gas Flammability limit vol% Flammability limit vol%

Lower Upper

Hydrogen 4.0 75.0

Methane 5.0 15.0

Propane 2.1 9.5

Acetylene 2.5 100.0

2.1.5 Thermodynamic properties

Heat capacity describes the amount of heat required to change the temperature of a substance. For an ideal gas, Mayer’s relation between the specific heat at constant pressure and the specific heat at constant volume is derived as:

(23)

denoting C_p as the specific heat capacity at constant pressure, C_v as the specific heat capacity at constant volume andR_sthe specific gas constant [17].

C_p can be expressed empirically allowing adjustments as the temperature changes. Usually, the molar heat capacities are expressed as a five-term polynomial of fourth order. Used in FLACS, Arntzen [20] simplified the polynomials into a second degree polynomial, at constant pressure becoming:

C_p,i =a_i+b_iT, (2.8)

citing a and b as specific heat capacity coefficients of temperature and are unique for each gas, i. Tab. 2.4 contains values for a and b for hydrogen, nitrogen, oxygen and water.

Table 2.4: Specific heat capacity coefficients for various gases.

Gas a_i b_i

H2 13600 1.719

N2 950 0.112

O2 1036 0.118 H2O(l) 4000 0.550

A ratio describing the heat capacity at constant pressure to heat capacity at constant volume is denoted by:

γ = C_p Cv

. (2.9)

2.2 Laminar burning velocities

The propagation rate, commonly called the burning velocity for any fuel is an important parameter tool, both as an input parameter and also for validation of combustion kinetics simulations. A one-dimensional combustion model can be derived for the laminar burning velocity describing the unburned gas-air mixtures being absorbed by the combustion reaction. Making measurements on real flames and transforming them, required one-dimensional values not present in nature can be obtained. Idealized adiabatically with no heat loss, no buoyancy and no interference by the wall, a planar laminar combustion at constant pressure has been illustrated by Eckhoff [18], see Fig. 2.1. If the gas mixture is ignited in the open end of the tube (Fig. 2.1a), the

(24)

combustion products will expand freely into the ambient atmosphere, and the observed flame speed S_f will be the same as the laminar burning velocity S_u. The distinction between the observed absolute flame speed and the laminar burning velocity is of vital importance as the latter is only defined relative to a fixed reference frame. Shown in Fig. 2.1b, ignition in the closed end of the tube will cause the expansion of the combustion produced to occur in the same direction as that of the flame propagation, with the burned gas generated behind being stationary. Therefore the observed flame speed Sf, in relation to the tube wall, will be the sum of S_u and the unburned gas flow velocity S_g.

Figure 2.1: One-dimensional combustion model with premixed fuel gas-air burning at constant pressure with a plane, laminar flame in a tube [18].

Similarly, in spherically expanding flame as shown in Fig. 2.2, the absolute flame speed S_f can be observed directly measuring the velocity from the center as the flame expands, i.e change in the radius with respect to time:

Sf = dr_f

dt , (2.10)

wheredrf is the change in radius, and dt change in time.

(25)

Figure 2.2: Ideal laminar spherical burning and expansion of a premixed gas/air mixture, after ignition at midpoint [18].

It can be further compared to the linear one-dimensional case, whereby ignition occurs in the closed end, that the spherically expanding flame is a combination of the burning velocity and the unburned gas flow velocity.

Assuming a very thin flame and negligible buoyancy, the unburned gas mixture will always be pushed in the direction of the propagation. The relationship between the absolute flame speed S_f and the laminar burning velocityS_u can be shown as:

Sf = drf

dt = ρu

ρ_b ·Su =E·Su, (2.11) defining E as the expansion ratio with ρ_u andρ_b being the gas density of the reaction, unburned and burned respectively. [18].

2.3 Flame Stretch

A flame front propagating in a non-uniform flow is subject to strain and curvature effects, and these changes in notion are measured by flame stretch.

Defined by Poinsot and Veynante [21], flame stretch is a fractional rate of change of a flame surface element and can be expressed as:

α= 1 A

dA

dt , (2.12)

(26)

where A is flame surface area. Writing the stretch rate in terms of the radius of the sphere gives:

α = 1 A

dA dt = 2

r_f dr_f

dt , (2.13)

and finally expressed as a function of the absolute flame speed by combining with Eq. (2.10):

α= 2

r_fS_f, (2.14)

presenting the total flame stretch as the sum of stretch caused by the flow non-uniformity and the stretch caused by the curvature found locally in the reaction front.

2.4 Estimating burning velocities

Flame speed whereby all speeds can be unambiguously defined and measured is rarely the case, as flames are subject to stretch effects of varying degree due to thermo-diffusive and hydrodynamic nature within and ahead of the flame.

This makes it more difficult to evaluate both numerically and experimentally.

At an early stage of a spherical flame propagation, the flame stretch effects are considerable and the pressure rise negligible. Later however, the pressure rise rate increases greatly and the stretch becomes negligible. These two behaviours describes the two different methodologies in extracting the burning velocity, with the former being the constant-pressure method, and the latter the constant-volume method.

2.4.1 Constant-Pressure Method (CPM)

The linear extrapolation methodology

Studies reiterated by Poisont and Veynante [21] suggests that in the limit of small strain and curvature terms, stretch (α) is the only parameter controlling

(27)

the flame structure and therefore, the displacement and consumption speeds through a linear relationship:

Sfs =Sfu−L·α, (2.15)

defining L as the Markstein length, S_f_s as the flame speed and S_f_u the value of the unstretched flame speed. The Markstein length describes the linear relationship between the stretch of the flame, either strain or curvature, and the flame speed. In other words, it is a measure of of how much the stretch of the flame is influenced by the flame speed - the higher the Markstein length, the larger the effect of stretch of the flame.

The non-linear extrapolation methodology

Through the work of Kelley and Law [22], a consideration of non-linearity between the flame speed and the flame stretch rate is determined. Restricting their analysis to "flames that are adiabatic and propagate in a quasi-steady manner", it is expressed as:

Sfs

S_f_u 2

ln Sfs

S_f_u 2

=−2L·α

S_f_u . (2.16)

The non-linearity was confronted in the flame response as a consequence of the small diffusivity of heavier fuels, casting considerable uncertainty on the feasibility and accuracy of the conventional method of linear extrapolation.

Both the linear and non-linear extrapolation method will be utilised to optimize estimation of the laminar burning velocity using the constant-pressure method.

2.4.2 Constant-Volume Method (CVM)

The dynamic pressure-time history in a closed vessel explosion can be used to obtain the laminar burning velocity through different relations. The validity of the pressure-time history approach is according to Omari and Tartakovsky [23]

based on the following assumptions in addition to ideal gas behaviour and uniform pressure distribution:

(28)

• The confinement acts against the expanding burned gas and hence, (S_f) is not equal to the propagating flame outwards, but a combination of both the unburned gas flow velocity and laminar burning velocity

• The confinement implies a pressure rise which is followed by a temperature rise due to compression. Thus, the burned and unburned gas temperature, as well as its density, continuously increases during flame propagation.

• In the analytical confined flame model, the flame stretch is neglected.

This is justified by realizing that during the pressure rise period, r_f is large and the flame front speed (dr_f/dt) strongly reduces, thus contributing to a continuously decreasing stretch rate, α.

The first expression consisting of both the dynamic pressure rise and optical accessed flame radius is known as the Fiock and Marvin expression [24] and is given as:

S_u = dr_f

dt − R²_v−r_f³ 3γ_ur²_f

1 p

dp

dt, (2.17)

where pis the dynamic pressure; γ_u is the specific heat ratio of the unburned gas; R_v is the radius of the vessel. The first term represents the absolute flame speed, S_f, and the second term representing the unburned gas flow velocity through the pressure rise inside the vessel, S_g.

A second and more widely used expression, only utilising the dynamic pressure rise is possible. To do so, the mass fraction x related to the pressure rise is found corresponding to the radius of the flame as:

r_f R_v =

"

1− p

p_i

−1/γu

(1−x(p))

#1/3

. (2.18)

When using the confined flame model equation and considering x to be obtainable from pressure monitoring i.e. x = x(p), the laminar burning velocity at elevated temperature and pressure, S_u is extracted from the pressure rise using the following equation:

S_u = R_v 3

dx dp

dp dt

p p_i

−1/γu"

1− p

p_i

−1/γu

(1−x(p))

#−2/3

, (2.19)

(29)

The derivation itself is from O’Donovan and Rallis [25] and is described in detail in their work.

Relating the burned mass fraction x to the pressure rise during a constant volume, taking into account the temperature rise in both the burned gas and the unburned gas zone, the analytical x−p relation proposed by Luijten et al. [26] considers energy conservation in the whole combustion vessel and is as follows:

x= p−p_i·f(p)

p_max−p_i·f(p), (2.20) with f(p) being:

f(p) =

γ_b−1 γ_u−1

+

γ_u−γ_b γ_u−1

p p_i

^γu_γu⁻¹

. (2.21)

p_max is the final pressure defining the final maximum pressure obtained when all of the gas is burned. Differentiating is straightforward, yielding:

dx

dp = 1−pif⁰(p)

p_max−p_if(p)+ pif⁰(p)[p−pif(p)]

[p_max−p_if(p)]² , (2.22) where,

p_if⁰(p) =

γ_u−γ_b γ_u

p p_i

−1/γu

. (2.23)

Both Eq. (2.20) and Eq. (2.22) are inserted into the the differential Eq.

(2.19), and evaluated computationally using MATLAB. The end result is not written explicitly as it is not considered necessary.

Once S_u is extracted, the value for S_u₀, defined at reference conditions of pressure as the one-dimensional laminar burning velocity, can be established.

The following equation is by Omari and Tartakovsky [23]:

Su Su0

= T_u

Ti

a p pi

b

, (2.24)

(30)

For a constant-volume adiabatic combustion, the unburned gas temperature rises due to isentropic compression and hence, Eq. (2.24) can be simplified into:

Su S_u₀ =

p p_i

α

, (2.25)

where α=a((γ_u−1)/γ_u) +b. Exponents a andb are weak functions of the equivalence ratios, and these values in addition to S_u₀ are fitted accordingly for each mixture.

2.5 Schlieren imaging technique

Air as we see it is experienced transparent with it propagating homogeneously through around us. Looking at other phases in likewise transparent form such as water and ice, one might see a reflection of the image around and even more importantly, a refraction of the transparent media showing the background.

Air may be without reflection, but it does have very weak refractive indices and although invisible to the eye, can be apparent with the schlieren imaging technique.

The basic concept of a schlieren imaging system, is to translate phase fluctua- tions into a visual optical image. In aerodynamics, flow visualization can be conducted in which changes in the index of refraction due to variation of the flow density, pressure or temperature can be measured. To apply quantitative analysis to schlieren optics systems, the physics behind are outlined with guidance of the work of Settles, Mazumdar and Lien et al. [27–29].

2.5.1 Light propagation

The refractive index, as light interacts with matter is given as

n=c/v. (2.26)

This indicates change through a transparent medium, with cbeing the universal speed of light in a vacuum, 3·10⁸ m/s, and v the light speed in the

(31)

medium. In the case of air and other gases, there is a simple relation between the refraction index and the gas density (ρ) showing:

n−1 =kρ, (2.27)

that puts the density in relation to n and k, the Gladstone-Dale coefficient, which is specific for every gas and its conditions, and varies between 0.1 and 1.5 cm³/g.

A simple interpretation of schlieren light refraction is best imagined in the light of a x, y, z-coordinate system. In a simple 2D-case, a planar light wave along the z-axis becomes displaced after propagating through an area of optical inhomogeneities in proportion to the refractive index. As shown in Fig. 2.3, light initially vertical becomes bent over a differential distance in

∆z/∆t.

Figure 2.3: Diagram of light rays entering a new material and being refracted in proportion to a refractive-index gradient [28].

With a differential angle ∆ξ, the refractive index from Eq. (2.26) can be rewritten to give an en expression of the angle:

∆ξ = c/n₂−c/n₁

∆y ∆t. (2.28)

(32)

The differential time ∆t can be expressed by the differential distance ∆z and the local velocity v =c/n, and by simplifying leads to:

∆ξ = n n₁n₂

(n₁−n₂)

∆y ∆z. (2.29)

The term ⁿ/n1n2 can be simplified to 1/n in the limit as∆y approaches zero.

Letting all finite differences approach zero yields:

dξ dz = 1

n dn

dy. (2.30)

Implementing the equation above for a more general case where other refractive index gradients may occur, it is possible to postulate that the angleξis equal to dy/dz. Enforcing this and turning from total derivatives to partial derivatives to account for the general case, we may obtain for the ray curvature in the y-direction:

∂²y

∂z² = 1 n

∂n

∂y. (2.31)

In similar terms, the ray curvature in the x-direction may be given:

∂²x

∂z² = 1 n

∂n

∂x. (2.32)

Eq. (2.31) and (2.32) both indicate light defections bending towards areas of higher refractive index. That is, the larger the refractive index gradient is, the larger the angle of refraction.

2.5.2 The z-type two mirror schlieren system

When setting up a standard schlieren system there are two major categories for collimating and focusing the light: convex lenses and parabolic mirrors [30].

The two groups differ very little optically, although using mirrors has become preferred as the most used approach as quality of parabolic mirrors has improved substantially [31]. Furthermore, for larger flow visualisation, mirror set-ups are favourable, and will be the subject of this section.

(33)

Using two identical parabolic mirrors on each side of the test area, in addition to a light source, a knife-edge and a capturing device, a z-type schlieren system may be achieved. Setting up the components in a single zigzag-line that suggests the letter z, its name is inspired by the distinguishable set-up.

As seen with a simple sketch in Fig. 2.4, the light source is placed with a certain angle θ sending a collimated beam of light to the first parabolic field mirror. A parallel beam between the mirrors through the test area reaches the second mirror. The second mirror with the same angle θ on the opposite side of the centerline is pointed towards a knife-edge.

Figure 2.4: Z-type schlieren arrangement [27].

The task of the knife-edge, positioned at the focal point of the mirror, is to block a portion of the incoming light. This allows the unblocked light to show gradients in the light intensity depending on the refractive index. In the test area, higher density gradients will be visible by brightened and/or darkened intensity in the pictures compared to the background. An example of this is shown in Fig. 2.5, as the density contrast in the explosion products propagation.

(34)

Figure 2.5: Schlieren image showing density gradients in the test area after a hydrogen-air explosion.

(35)

Chapter 3 Previous work

3.1 Laminar burning velocities of hydrogen-air mixtures

Experimentally, quantitative investigation and determination of gaseous mixture combustions have been developed and upgraded since the first recorded estimation of a burning rate of a methane-air flame published in 1815 by Sir Humphrey Davy [32]. In the period up until today, the definition of burning velocity has evolved as the experimental work has progressed. Taking into account of the effect of stretch on propagating flames, a shift in the techniques used ensued. Stabilized (stationary) measurement technique, such as flat flame burners and counter double flame were exchanged for a spherical expanding flame technique. It was after a critical review done by Andrews and Bradly [33] which considered it to be the best method for measuring propagating flame velocity, although in present times it is still an ongoing dis- cussion. A research paper by Gelfand et al. [34] collected data illustrating the trend of different methods and corresponding results of hydrogen-air mixture, presented in Tab. 3.1. Even when the method applied was alike, discrepancies still occurred in S_u^st and S_u^max, implying that there are still uncertainties present due to radiation, stretch effects, confinement and preferential mass diffusion effects [23].

Something in common with the spherical expanding flame techniques is the complexity in data processing and needed requirements in equipment such as a high-speed camera and parabolic mirrors. Therefore, a labour and cost-effective burning velocity measuring method was convenient using a

(36)

Table 3.1: Data by various authors showing the hydrogen-air premixed laminar burning velocity measured at stoichiometry, S_u^st, and maximum, S_u^max, using different methods [34].

(37)

confined method based on the pressure-time history approach, with perceivable decreased accuracy [35]. A study done by Dahoe [14], used a 169 ml vessel and compared laminar burning velocities obtained with other laminar burning velocities which concluded in unison results. Illustrated in Fig. 3.1, it was seen that data obtained using CVM (straight line) were seen to fall within the scatter of data obtained by more advanced methods fully taking into account the influence of flame stretch.

Figure 3.1: The comparison of laminar flame velocities of hydrogen-air mixtures measured by various techniques [14].

3.2 Laminar burning velocities of hydrogen-air mixtures exposed to water mist

Dividing water into fine droplets gives it new properties and capabilities, which has constantly been explored since the mid 1950s [36]. Measuring the effectiveness of an inhibiting agent such as water mist can be done by examining the laminar burning velocity as an indicator. It indicates the relative effectiveness of suppressants in isolation from other factors affecting burning velocity, also noting that the reduction in burning velocity is not simply proportional to the change in heat capacity of the mixtures [37].

Experiments on the behaviour of premixed hydrogen-air mixtures exposed to water mist has been conducted by Cheikhravat et al. [38] using a 56 litre spherical vessel at atmospheric conditions. Visualization of the flame by

(38)

schlieren imaging technique made it possible to investigate the interaction of a flame with water droplets, as shown in Fig. 3.2. Focusing on the water droplets’ characteristics, it was found that the droplet effect on the flame depends strongly on the droplet diameter. To quench the flame efficiently, the droplet must be totally evaporated during the residence time in the flame front thickness. A Sauter Mean Diameter (SMD) ≈ 5µm reduced the burning velocity over a wide range of equivalence ratios φ = 0.6-2.9, and also slightly mitigating the final maximum pressure pmax. Relatively large droplets (SMD = 200-250µm) was also investigated showing mitigating effect to the evolution of the combustion.

Figure 3.2: Ignition and propagation of hydrogen-air-mist flame fed through a nozzle initially at 305 K and 1 bar without(a) and with(b) water mist [38].

van Wingerden et al. [39] exhibited laboratory-scale tests done with methane- air mixtures proving water droplets maximum effectiveness in mitigation laminar burning velocities in the order of 10 µm. These droplets had then the capability of evaporating fully in a laminar flame, established using Fig. 3.3.

To achieve full evaporation, it was approximated a flame thickness of 1 mm and a residence time of the droplet in the flame of 2 ms. For higher relative velocities, such as in a hydrogen-air combustion, the evaporation time would decrease as well as the residence time. Therefore it could also be concluded that at higher relative velocities, only droplets smaller than 10 µm would fully evaporate.

(39)

Figure 3.3: Evaporation time as a function of droplet diameter for various relative hot gas flow speeds around the droplet [39].

Studies by Holborn et al. [40–42], both experimentally and by modelling, showcased that fine water mist (5-6 µm) reduced the estimated burning velocity for hydrogen-air mixtures significantly. Experimentally using the pressure-time history method was applied to a large number of test data sets.

It confirmed that high concentration water mist fogs (0.2-0.3 kg/m³) reduced the laminar burning velocity for both lean and rich hydrogen-air mixtures.

It was also deducted that the peak overpressure for lower concentrations of hydrogen-air mixtures up to 20 vol% (φ <0.6) could be significantly reduced, but would require very high fog densities when the highest rate of pressure rise is present near stoichiometric mixtures.

Not all instances of water mist fog introduction result in lower laminar burning velocities. Geometry dependencies are also to be accounted for, when water spray is introduced into large volumes. Investigation into turbulence generation by water-spray systems has been very limited, especially in a hydrogen-air atmosphere prior to ignition. van Wingerden and Wilkins [43]

addressed the influence of turbulence generated by these systems on the course of gas explosion. Quantitatively, they found a burning rate increase factor from 1.5 to 2 for propane and 1.4 to 2.3 for methane, with the main source of turbulence relating to the bulk flow of water mist into the vessel. The droplet size or other parameters related to the droplet velocity had minor influence.

(40)

Chapter 4 Experimental set-up and procedure

This section describes the experimental equipment, set-up and general procedure. Essential equipment such as the high-speed camera and explosion chamber were already available in the lab for usage. Other equipment such as the data acquisition module and water cylinder were borrowed from Gex- con. The set-up and general procedure described were the established final arrangement and approach.

4.1 Experimental rig

4.1.1 Explosion chamber

The explosion chamber used for the work described in this thesis was a chamber previously used by Halland [15] and built by Skjold [44], where a detailed overview of the chamber can be found. For optical access to the explosion, there were two circular windows mounted on opposite sides of the sides for free sight through the chamber. Other holes were installed for inlet and outlet of gases, and for pressure sensors. For this work, an additional hole at the bottom of the chamber was made for inlet of water mist and its dispersion. Before the aforementioned hole was made, the total volume measured by water filling proved to be 20 235 cm³ with the dimensions being 27.3 cm x 27.3 cm x 26.7 cm. A test carried out certified that the chamber could cope with pressures up to 20 barg. A general set-up of the explosion

(41)

chamber and the equipment connected to it is presented in Fig. 4.1.

Figure 4.1: 20-litre cubical explosion chamber.

4.1.2 Windows

The windows mounted to the chamber consisted of two disks of natural fused quartz glasses, with a thickness of 51 mm and a diameter of 144 mm. Being of high quality, the glasses provided an unblemished background for light to propagate through. Irregularities inevitably present in or on the surface of the low-quality glass, be it small cracks, scratches or fragments, may cause a minor deflection on the light and sabotage the quality of the schlieren images.

4.1.3 Piezoelectric pressure sensor

A quartz pressure sensor type Kistler 7031 with a pressure range up from 0-250 bar was utilized for measuring the dynamic pressure rise throughout the explosion with a frequency range up until 80 kHz. The built-in accelerometer compensates interference signals produced by shock or vibration in the direction of the sensor axis acting through the diaphragm on the quartz crystal measuring element. The signal is then transformed from pressure into an electric charge.

(42)

The pressure sensor was mounted in the explosion chamber wall with a Kistler 7501 mounting adapter, sealing off and impeding external liquids from entering into the connector and corrupting the sensors electrical isolation.

Finally a Kistler 7401 BNC-mounting nipple was connected to the adapter attaching the pressure sensor to the charge meter, a Kistler 5015. The sensor, adapter and nipple is presented in Fig. 4.2. The charge meter connects to the computer displaying instantaneous, peak and average values converting the electric charge back to pressure values. The cable to remotely control the charge amplifier was not available in the laboratory at the time. Therefore, activating the charge amplifier had to be done manually just before starting the triggering sequence to reduce the certain drifting of the pressure sensor.

Figure 4.2: From left to right - Piezoelectric pressure sensor (Kistler 7031);

Mounting adapter (Kistler 7501); BNC-mounting nipple (Kistler 7401).

4.1.4 Ignition system

A capacitive spark generated in the gap between two horizontal electrodes centred in the chamber ignited the flammable mixture in the chamber. The electrodes were connected to a spark generator made and modified by Werner Olsen, a senior engineer in electronics employed at UoB. The energy released by the spark generator varies depending on the type of electrodes used and the spark-gap length. In this work, sharpened wolfram electrodes were used with a thickness of 2 mm. The spark gap was adjustable and was usually fixed at ≈ 2 mm, giving the distance for a successive energy release.

(43)

Energy stored in a capacitive spark can be calculated by:

E = 1

2CU², (4.1)

where C is the capacitance and U is the voltage.

The capacitor used had a voltage of 362 V and a capacitance of 0.2 µF, and by using Eq. (4.1) gave an energy of 13.1 mJ. Energy losses in radiation, to the electrodes, because of skin effect resistance reduces the theoretical calculated energy by 60-90%, concluded in a study done by Randeberg et al [45]. Werner Olsen, one of the authors, with additional regards to the spark-gap approximated an energy loss of up to 75%, therefore giving a more correct value for the effective energy released to be 3.27 mJ. This value was significantly higher than the minimum ignition energy measured by Ono et al. [46] to be only 0.019 mJ. This was to ensure ignition independent of mixtures, spark gap and existence of different phases when water mist is present. A more accurate method of measuring the actual energy released with a oscilloscope was attempted, but was later sidelined due to complications in signal observations.

4.1.5 Vacuum and gas delivery system

A N86 KTP Laboport vacuum pump was used to evacuate the contents of the explosion chamber between each test at a nominal flow rate of 5.5 L/min.

The mixture preparation was performed by following Dalton’s law of partial pressures, stating that in a mixture of non-reacting gas, the total pressure exerted is the sum of the partial pressures of the individual gasses. A real-time value of the internal chamber pressure was given by a Druck DPI 705 IS (-1 to 1 bar range) hand-held manometer with a 0.1 mbar gauge readout resolution with a full scale accuracy of 0.1%. Hydrogen of 99.9999 % purity and compressed air was introduced into the chamber by means of needle valve control, allowing for fine adjustment of the internal partial pressures.

(44)

4.2 The schlieren optical system

4.2.1 Field mirrors

Two spherical field mirrors were used as part of thez-type alignment. Both mirrors were identical with 150 mm in diameter(D) and with focal length(f) of 1500 mm. The focal length is a measure of light convergence. Any mirror can be characterized based on their diameter and focal length by the following equation describing the f/number:

f/number = f

D. (4.2)

The f/number for the mirrors used in this work describing the speed and clarity of the optical system was f/10. The lower the f/number, the wider, or bigger the aperture. Both mirrors were attached to metal sticks mounted to a metal beam. The sticks had lines engraved for easy adjustment and integrating the same height.

4.2.2 Light source, lens and slit

The light source was a single Light-Emitting Diode (LED) lamp with separated battery holder and charger, demounted from a rechargeable head torch model MF-H05. It has two power modes - high power with a capacity of 350 lumen and low power with a capacity of 180 lumen, with the former mode being used.

The use of a LED lamp instead of a light source directly connected to a wall socket is related to the current delivered. Power sockets provide an alternate current, interchanging its direction with a 50 Hz frequency, which is little favourable when filming in high-speed. A LED-lamp with batteries however delivers a direct current, flowing only in one direction thereby without any fluctuation in the light on view.

The divergent beam from the LED-lamp was fixated on a condenser lens with aim to render the beam converging it towards a circular slit. The lens has a diameter of 90 mm with a focal length of 150 mm. An iris diaphragm acted as a slit with an opening at its center, with an adjustable opening in the diaphragm thus varying the amount of light let through the slit opening. It was placed at the focal point of the condenser lens and functioned as the point light source towards the first mirror. The lamp mounted focused on

(45)

the condenser lens, with the converging beam focused on the slit acting as the point light source is presented in Fig. 4.3.

Figure 4.3: Light, lens and slit aligned accordingly.

4.2.3 Knife-edge

A knife-edge was used as a cut-off of the refracted light towards the camera. The knife-edge was effectively a utility knife, mounted vertically on a tripod. Considering the positioning tweaks a knife-edge required, a tripod was favourable for its adjustability. Positioning of the knife-edge plays a significant role in the quality of the schlieren image recorded. Demonstrated in Fig. 4.4, a correctly positioned knife-edge will uniformly darken the image.

(46)

This is opposed to if the knife edge is too close or too far away from the second mirror.

Knife-edge too close Uniform darkning Knife-edge too far

Figure 4.4: Illustration of the effect of the knife-edge movement on the schlieren image for a vertical knife edge cutting parts of the light beam.

4.2.4 High-speed camera

The high-speed camera used in the schlieren system was a Photron Fastcam SA4 RV, and could produce 3 600 frames per second (fps) with 1024x1024 pixel resolution, and operate frame rates up to 500 000 fps at a much reduced resolution. For the experiments conducted in this work, a frame rate of 20 000 fps with a corresponding resolution of 512x325 pixels was chosen as a middle ground between resolution and frame rate. The images were initially stored in the camera unit and then transferred to a computer before being processed.

The high-speed camera control, image and video editing and download software, Photron FASTCAM Viewer (PFV), was used. It enabled easy set up of the camera as well as adjustments of the frame rate, resolution and shutter- speed. Mounted to the high-speed camera was a Tamron macro lens with a focus ring clutch-type manual focusing and had a focal length from 70-200 mm with a fast F2.8 constant maximum aperture. For zooming purposes, a lens of this nature was essential in being able to magnify into the schlieren image and viewing details such as ignition spark and the instabilities in the spherically expanding flame.

(47)

4.3 Data acquisition and triggering system

For a systematic and proper routine during the test sequence, a data acquisition module, NI USB-6259, from National Instruments was utilized, see Fig. 4.5. It provided analog and digital triggering, and was optimized for superior accuracy at fast sampling rates. A corresponding program developed in LabVIEW by Gexcon was used to trigger all the required components using BNC-cables with a 5 V pulse; camera, charge amplifier, spark generator and solenoid valve.

Figure 4.5: Data acquisition module, NI USB-6259, used for equipment- triggering and acquisition of amplified pressure signals in pressure measurements.

The charge amplifier was utilized by the data acquisition module to designate the duration of the measure time of the pressure, and the measure frequency.

In this work, a frequency of 50 000 Hz was chosen to ensure sufficient pressure

Experimental Determination of Hydrogen-Air Laminar Burning Velocities, and the Effect of Water Mist